Face Aftereffects and the Perceptual Underpinnings of Gender-Related Biases

Recalibrating Gender Perception: Face Aftereffects and the Perceptual Underpinnings of Gender-Related Biases

David J. Lick and Kerri L. Johnson University of California, Los Angeles

Contemporary perceivers encounter highly gendered imagery in media, social networks, and the work- place. Perceivers also express strong interpersonal biases related to targets’ gendered appearances after mere glimpses at their faces. In the current studies, we explored adaptation to gendered facial features as a perceptual mechanism underlying these biases. In Study 1, brief visual exposure to highly gendered exemplars shifted perceptual norms for men’s and women’s faces. Studies 2–4 revealed that changes in perceptual norms were accompanied by notable shifts in social evaluations. Specifically, exposure to feminine phenotypes exacerbated biases against hypermasculine faces, whereas exposure to masculine phenotypes mitigated them. These findings replicated across multiple independent samples with diverse stimulus sets and outcome measures, revealing that perceptual gender norms are calibrated on the basis of recent visual encounters, with notable implications for downstream evaluations of others. As such, visual adaptation is a useful tool for understanding and altering social biases related to gendered facial features.

Keywords: visual aftereffects, face perception, social vision, gender bias, gendered phenotypes

In daily life, people encounter others who vary considerably in included many men with visibly feminized facial features (high their gendered appearance (i.e., masculinity/femininity). Although brow line, high cheekbones, wide eyes, small nose; e.g., Max many individuals fall within the average range of gender variation, Greenfield, Matt Bomer, Damian Lewis, Paul Rudd, Usher, Aaron others do not. In fact, self-presentational motives and strategic Paul, Ian Somerhalder). Individuals’ everyday self-presentations editing practices ensure that the people we see in print, onscreen, also favor feminized phenotypes, insofar as both men and women or in daily life are often exceptionally gendered. One might expect undergo cosmetic surgery to accentuate feminine aspects of their that such gendered imagery would exaggerate sexually dimorphic facial appearance, such as small noses and lifted eyelids (Davis, features (i.e., exceptionally feminine women and masculine men), 2002). Taken together, these factors produce an overrepresentation but that is not the case. Instead, a systematic feminization is of feminine phenotypes relative to more masculine phenotypes in evident across visual domains. At least in Western countries, media and social life. female actresses, reporters, politicians, and models tend to embody Such highly gendered imagery is likely to shape our preferences hyperfeminine features (Carpinella & Johnson, 2013; Carter & for and judgments of other people. Indeed, gendered facial features Steiner, 2004; Collins, 2011; Heldman & Wade, 2011). Perhaps predict biased impressions of men and women, with perceivers surprisingly, the same is true among Western men: Visual media expressing distaste for exceedingly masculine features in both rarely depict men who are hypermasculine, but increasingly depict sexes (Frederick & Haselton, 2007; Little & Perrett, 2011; Sc- men who are somewhat feminized (Adams, 2011; E. Anderson, zensy, Spreeman, & Stahlberg, 2006). We contend that frequent 2005, 2008, 2009; Coad, 2008). Anecdotal evidence corroborates visual exposure to feminine phenotypes, or infrequent exposure to these empirical observations. For example, People Magazine’s list more masculine phenotypes, may perpetuate these gender-related of the most prominent and attractive male celebrities of 2012 biases. Consistent with this possibility, empirical evidence suggests that visual exposure to faces helps to calibrate what is This document is copyrighted by the American Psychological Association or one of its allied publishers. perceived as normative and that shifts in perceptual norms may This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. influence subsequent social evaluations. For example, when per- This article was published Online First September 30, 2013. ceivers adapt to caricatured faces, features of those faces begin to David J. Lick, Department of Psychology, University of California, Los appear normative (Rhodes, Robbins, et al., 2005). Subsequent Angeles; Kerri L. Johnson, Department of Psychology and Department of faces that violate this norm (i.e., anticaricatures) tend to be eval- Communication Studies, University of California, Los Angeles. uated negatively, a shift in judgment known as an aftereffect This research was supported by a National Science Foundation Graduate (Principe & Langlois, 2012; Rhodes, Jeffery, Watson, Clifford, & Research Fellowship and Janet Hyde Graduate Student Research Grant Nakayama, 2003). Although compelling, evidence of this link (David J. Lick) as well as by National Science Foundation Grant BCS- between visual adaptation and social evaluation has been limited 1052896 (Kerri L. Johnson). We thank Jon Freeman for stimuli and members of the Social Communication Lab for assistance with data col- primarily to caricatured features and ratings of attractiveness. lection. Whether and how visual adaptation to naturally varying pheno- Correspondence concerning this article should be addressed to David J. types affects broader social evaluations remains less clear. Lick, UCLA Department of Psychology, 1285 Franz Hall, Box 951563, Here, we argue that visual adaptation to highly gendered phe- Los Angeles, CA 90095-1563. E-mail: [email protected] notypes influences gender-related biases in social perception. Spe-

1259 1260 LICK AND JOHNSON

cifically, we propose that people evaluate others on the basis of when those features become extreme (Rhodes, 2006; Rhodes et al., perceived norms for sex categories, such that targets whose gen- 2000). This may lead to more favorable evaluations of feminine dered appearances lie far from the norm are evaluated negatively. men relative to more masculine men. In line with this reasoning, Furthermore, we propose that these norms are malleable and that one study noted a curvilinear association between masculine ap- they are calibrated on the basis of the faces that observers have pearance and ratings of male attractiveness, such that women recently encountered in their environment. If our hypotheses are perceived masculine body shapes to be maximally attractive to a correct, then visual adaptation may be a proximal mechanism point, after which additional musculature was perceived as unat- underlying gender-related biases that arise in the early stages of tractive, perhaps due to its association with excessive strength social perception, and it may provide a novel tool for understand- (Frederick & Haselton, 2007). In other work, experimental admin- ing and eventually mitigating those biases. istration of oxytocin masked observers’ feelings of threat when viewing hypermasculine faces. This led observers who received Gender-Related Biases in Social Perception oxytocin to prefer more masculine faces relative to observers who did not receive oxytocin (Theodoridou, Rowe, Rogers, & Penton- People readily form impressions of others on the basis of mere Voak, 2011). Collectively, these findings suggest that social eval- glimpses of their faces (Willis & Todorov, 2006), and they often uations may be biased toward feminine men and women due to exploit gendered information to do so (Freeman, Johnson, Am- perceivers’ stereotypically positive associations with femininity bady, & Rule, 2010; Keating, 1985). Indeed, social evaluations and negative associations with masculinity. Although this expla- based on gendered appearance cues begin within milliseconds of nation is certainly reasonable, other factors may also be at play. In visual exposure (Ito & Cacioppo, 2000; Ito, Thompson, & Ca- particular, we propose that perceptual experience may contribute cioppo, 2004), and they carry implications for a wide range of to gender-related biases in social evaluation. social evaluations. For example, gendered features guide attractiveness judgments (Johnston & Franklin, 1993; Rhodes, Hickford, Perceptual Experience, Gender Norms, & Jeffery, 2000). This results in a pronounced bias favoring and Evaluative Biases feminine women: Across cultures, feminine female faces (i.e., those with a rounder jaw, wider eyes, higher brow line, higher Biases against hypermasculine men and women may occur in cheekbones, smaller nose) are consistently rated as more attractive part because perceivers have limited visual exposure to such than masculine female faces (Jones & Hill, 1993; Perrett et al., targets. Indeed, the women portrayed in popular media, politics, 1998; Perrett, May, & Yoshikawa, 1994; Rhodes et al., 2000; video games, and social networks tend to be more feminine than Valenzano, Mennucci, Tartarelli, & Cellerino, 2006). Gendered masculine (Carpinella & Johnson, 2013; Manago, Graham, Green- appearance cues also bias judgments of men’s attractiveness, field, & Salimkhan, 2008). A similar trend exists for men: Whereas though the direction of the effect is somewhat less clear. In some portrayals of hypermasculine men are rare, portrayals of more studies, perceivers judge masculine male faces (i.e., those with a feminine men are increasingly common (Adams, 2011; E. Ander- squarer jaw, smaller eyes, higher forehead, larger nose, and thicker son, 2005, 2008, 2009; Coad, 2008; Harris & Clayton, 2007; eyebrows) to be most attractive (Grammer & Thornhill, 1994; Swain, 2006). In fact, although it may seem counterintuitive, Johnston, Hagel, Franklin, Fink, & Grammer, 2001; Scheib, Gan- exposure to somewhat feminized men seems to have outpaced gestad, & Thornhill, 1999). In other studies, perceivers judge exposure to highly masculine men. Consider, for example, how feminine male faces to be most attractive (Little & Hancock, 2002; often one sees faces similar to Rambo or Arnold Schwarzeneg- Perrett et al., 1998; Rhodes et al., 2000). ger—exposure to these sorts of hypermasculine faces is rare. Researchers have obtained similar patterns of gender-related However, many of our friends, family members, politicians, and biases for social evaluations other than attractiveness. For exam- influential celebrities exhibit less masculine, and even somewhat ple, gendered phenotypes have been shown to predict diverse feminine, facial features. This is not to imply that contemporary social judgments, including a person’s respectability, sincerity, men exhibit the same femininity as women or that they embody prosociality, honesty, warmth, loyalty, likability, intelligence, and distinctly feminine behaviors and traits, but rather that they tend dependability (Lick & Johnson, 2013; Little, Jones, Penton-Voak, not to exhibit hypermasculine phenotypes. Consequently, we argue Burt, & Perrett, 2002; Penton-Voak et al., 2007; Principe & that perceivers have less perceptual exposure to hypermasculine This document is copyrighted by the American Psychological Association or one of its allied publishers. Langlois, 2012; Sczensy et al., 2006). The direction of these phenotypes relative to somewhat feminized phenotypes of both This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. effects is common to judgments of both men and women, such that sexes. feminine targets of both sexes receive more favorable evaluations To the extent that visual exposure is skewed toward feminine than their masculine counterparts. phenotypes, preferences for targets with feminine facial features Such mounting evidence of a preference for feminized features may emerge. Indeed, impression formation relies heavily on per- in both men and women has been surprising to many, and several ceptual experience, and researchers have long known that perceiv- research teams have begun to probe these findings in greater depth. ers prefer prototypical category exemplars with which they have One emergent explanation is that gendered phenotypes predict more experience to nonprototypical exemplars with which they evaluative biases via stereotypic associations. For example, in have less experience (Kahneman & Miller, 1986; Perry, 1994; women, feminine features are commonly associated with beauty, Posner & Keele, 1968). In fact, increasing exposure to a stimulus immunity, and fecundity (Symons, 1995; Thornhill & Gangestad, reliably increases preferences for that stimulus (Zajonc, 1968). 1996), which may lead to more favorable evaluations of feminine Such exposure effects extend to socially relevant evaluations as women relative to masculine women. In men, masculine features well: Merely observing outgroup members with whom one has had are associated with negative traits such as violence, especially limited interpersonal contact reduces prejudice against that out- FACE AFTEREFFECTS AND GENDER-RELATED BIASES 1261

group for up to 24 hr (Dasgupta & Greenwald, 2001; see also Mutz teristics with the adapting stimuli (Anzures, Mondloch, & Lackner, & Goldman, 2010; Schiappa, Gregg, & Hewes, 2005). 2009; Leopold & Bondar, 2005; Rhodes et al., 2003; Rhodes, Despite numerous studies linking perceptual experience to fa- Robbins, et al., 2005). These findings suggest that visual adapta- vorable evaluations, the mechanisms underlying these exposure tion shifts perceptual norms and evaluative biases in the same effects remain unclear. Recent work in social vision has offered a direction, such that perceivers rate faces that are similar to adapt- compelling hypothesis: Visual exposure may affect social biases ing faces as both more normative and more attractive. Further- by shifting perceptual norms for targets’ appearances. That is, more, a recent study revealed that visual adaptation leads to more stimuli may appear increasingly normative as perceivers gain favorable affective responses to human/chimpanzee facial morphs, additional exposure to them, leading to enhanced evaluative judg- as indicated by activity over the zygomaticus major muscle group ments. Studies of visual adaptation provide more direct support for (Principe & Langlois, 2012). Another study revealed that adapta- these claims. tion to masculine facial features enhanced the perceived trustworthiness of men’s faces (Buckingham et al., 2006); however, re- Adaptation Aftereffects searchers did not investigate more global evaluations or judgments related to women’s gendered features. Thus, extant research sug- Vision scientists have long recognized that perceptual norms are gests that adaptation to distorted facial features leads to more malleable and that relatively brief exposure to a class of stimuli favorable evaluations of similar faces. However, researchers have (adaptation) can alter norms, resulting in noteworthy perceptual yet to determine whether adaptation to naturally varying facial changes (aftereffects; Clifford & Rhodes, 2005; Webster, Werner, features (e.g., gender) reliably alters broad social evaluations re- & Field, 2005). Early researchers documented aftereffects follow- lated to those features. Such insights would help us to understand ing adaptation to low-level visual cues such as motion (Addams, the perceptual mechanics of gender-related biases in social per- 1834; Purkinje, 1825) and color (Hering, 1878). More recent ception. research uncovered aftereffects in higher level vision—most notably, in face perception (Leopold, Rhodes, Muller, & Jeffery, The Current Research 2005). Findings revealed that prolonged visual exposure to faces We propose that perceivers adapt to gendered features in men’s with a particular feature biases perception of subsequently encoun- and women’s faces and that these adaptations alter the perceptual tered faces in the opposite direction. For example, adaptation to norms that guide social evaluations. Specifically, we predict that large noses makes an average nose appear quite small. The dom- brief visual exposure to gendered facial features will shift percep- inant explanation for these aftereffects is that facial features are tual norms, such that perceivers who are exposed to masculine features coded relative to a perceptual norm that shifts in response to visual will view more masculine features as normative, whereas perceivers exposure (Rhodes, Jeffery, Clifford, & Leopold, 2007; Rhodes, who are exposed to feminine features will view more feminine Robbins, et al., 2005). That is, perceivers possess norms that code features as normative. We also predict that social evaluations will for particular features (e.g., the average nose in the population), vary concomitantly with these shifts in perceptual norms, such that and these norms are dynamically calibrated to accommodate recent perceivers who are exposed to masculine features will evaluate visual experiences. After exposure to large noses, an average nose masculine targets more favorably, whereas perceivers who are appears quite small because the perceptual norm for noses has exposed to feminine features will evaluate feminine targets more updated to appear bigger than it was to begin with. Thus, in favorably. If we are correct, then perceivers may express bias general, research on face aftereffects has revealed that visual against hypermasculine faces because they are exposed to them adaptation biases perception away from an adapting stimulus (i.e., less often than feminine faces, making them appear nonnormative exposure to large noses makes average noses appear small) by and therefore leading to negative evaluations. Experimental expo- shifting the norm toward the adapting stimulus (i.e., exposure to sure to masculine faces should mitigate these biases, whereas large noses makes large noses appear normative). exposure to feminine faces should exacerbate them. Face aftereffects are well documented for adaptation to physi- We tested the evaluative implications of visual exposure to cally distorted features (e.g., caricatures and anticaricatures; N. D. gendered facial cues in a series of related experiments. In Study 1, Anderson & Wilson, 2005; Fang, Ijichi, & He, 2007; Yamashita, we tested whether perceptual norms for men’s and women’s facial Hardy, De Valois, & Webster, 2005). In daily life, however, faces This document is copyrighted by the American Psychological Association or one of its allied publishers. features shifted following exposure to gendered cues in computer- vary in more phenotypically natural ways (e.g., pigmentation, This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. generated faces. In Study 2, we explored the evaluative implica- texture), and adaptation to naturally varying facial characteristics tions of such visual adaptation. In Study 3, we pinpointed the is poorly understood. The few studies in which adaptation to direction of these shifts in evaluation by including a control group. natural phenotypes has been examined revealed that such adapta- Finally, in Study 4, we tested the generality of these effects using tion does occur (Little et al., 2002; Webster, Kaping, Mizokami, & a different set of evaluative judgments and facial photographs of Duhamel, 2004). However, the downstream consequences of such real men and women that varied in their gendered appearance. adaptations remain unclear because most of the relevant studies Collectively, these studies support the use of visual adaptation as examined low-level face aftereffects such as perceived sex cate- a tool for understanding and potentially altering gender-related gory boundaries, gaze direction, and facial viewpoint (N. D. An- biases in social perception. derson & Wilson, 2005; Bestelmeyer et al., 2008; Fang et al., 2007). Study 1 Some evidence suggests that visual adaptation may alter higher level evaluations of faces. For example, visual adaptation results in Previous studies have demonstrated that visual adaptation alters more favorable attractiveness ratings for faces that share charac- perceived norms for faces, especially those with caricatured fea- 1262 LICK AND JOHNSON

tures (Rhodes, Robbins, et al., 2005; Webster et al., 2004). In of all 21 male or female faces that varied in gendered appearance Study 1, we sought to extend this work by testing whether adap- (Pretest). Next, participants were randomly assigned to one of two tation to gendered facial features alters perceptual norms for men adaptation conditions in which they viewed either the five most and women. On the basis of empirical and anecdotal evidence that feminine faces (Feminine Adaptation) or the five most masculine feminine individuals are more common than more masculine in- faces (Masculine Adaptation) from the sex category to which they dividuals in media and everyday social networks, we predicted that were assigned. These adaptation images were selected from the participants would enter the study with slightly feminized norms images in the array, and they were repeatedly displayed for 3 s for male and female faces. However, we expected that adaptation each in random order for a total of 3 min.2 Following adaptation, would alter these norms, such that exposure to masculine pheno- participants again identified the most average-looking man or types would lead participants to identify more masculine faces as woman from an array of all 21 male or female faces (Posttest). The normative, whereas exposure to feminine phenotypes would lead placement of faces in the array differed randomly at pretest and participants to identify more feminine faces as normative. posttest. Finally, participants completed the Personal Attributes Questionnaire (PAQ; Spence, Helmreich, & Stapp, 1975) to assess Method their endorsement of masculine and feminine gender schemas, and they reported their sex, age, race, and education before being Participants. Two hundred fifteen Internet users from the debriefed. United States (85 men, 124 women, six unreported) completed an In total, 106 participants evaluated male faces (52 masculine online study. Participants were diverse in terms of age (M ϭ 34.16 adaptation, 54 feminine adaptation) and 109 participants evaluated years, SD ϭ 13.42 years) and racial identity (78% White, 8% female faces (66 masculine adaptation, 43 feminine adaptation). Asian, 5% Black, 6% Latino, 5% biracial), and most reported a Participants were randomly assigned to adaptation conditions, so high level of education (90% attended college). we suspect that differences in sample size reflect differential Stimuli. Stimuli were 21 computer-generated male faces and dropout rates across conditions. A chi-square test revealed that the 21 computer-generated female faces that varied systematically in number of participants did not differ significantly across these four their gendered features (400 ϫ 477 pixels at 72 pixels/inch; see cells, ␹2(1) ϭ 2.87, p ϭ .09. Figure 1). We created the stimuli using FaceGen Modeler, which estimates phenotypic features based on parameters observed in Results and Discussion several hundred three-dimensional face scans of the human pop- Our primary aim in Study 1 was to explore participants’ per- ulation (Blanz & Vetter, 1999). Specifically, we began with Face- ceptual gender norms for faces and to test whether visual adapta- Gen’s average base face and set all phenotypic features (e.g., age) tion altered these norms. We predicted that norms would be notably at their anthropometric mean. We then used the gender-morphing feminized at pretest but that they would shift in the direction of adapta- tool to alter the features of the base face. For female faces, we tion, such that observers would find masculine faces more norma- systematically changed the apparent gender from the most femi- tive after exposure to masculine features but feminine faces more nine face possible to the point at which the face had equally male normative after exposure to feminine features. and female characteristics, yielding 21 female faces that varied Because our stimuli varied incrementally in gender typicality, incrementally in gender typicality from hyperfeminine to hyper- we numerically coded each face from 1 (hypermasculine)to21 masculine. For male faces, we systematically changed the apparent (hyperfeminine) to quantify the gendered features that participants gender from the most masculine face possible to the point at which perceived as average (Perceptual Gender Norm). Because partic- the face had equally male and female characteristics, yielding 21 ipants chose the most average face at two time points, we used male faces that varied incrementally in gender typicality from hypermasculine to hyperfeminine. Thus, our stimuli captured the full range of gendered variation for each sex, up to the point where 1 In this and all forthcoming studies, separate samples of participants the features were anthropometrically androgynous. We considered evaluated male and female targets. We conducted our studies in this way androgynous faces to be gender-atypical within a given sex be- because, in our experience, it is more efficient to collect several small samples rather than one large sample on Mechanical Turk. On the advice cause we explicitly informed participants of the sex category that of two anonymous reviewers, we have combined the samples and exam- they were judging (i.e., participants judging female faces were told ined Target Sex as a between-subjects factor to streamline the results. This This document is copyrighted by the American Psychological Association or one of its allied publishers. that would only see female faces). Therefore, hypermasculine analytic decision was justified insofar as there were no notable differences This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. women and hyperfeminine men were extremely gendered relative between the samples: Frequencies and means of all demographic variables were not significantly different across the samples evaluating men and to a sex-specific norm. To ensure that the stimuli were externally women. Still, it is important to note that we did not randomly allocate valid, we allowed all other facial features to covary with gender as participants to evaluate male and female faces. Although we did not find they do in the population. For example, masculine faces had darker any significant effects of Target Sex across studies, future research should skin tone than did feminine faces, because pigmentation naturally confirm these findings after randomly assigning participants to judge varies along gendered lines (see Johnson, Freeman, & Pauker, different sex categories. 2 To ensure that participants were engaged throughout the adaptation 2012). period, we informed them that we were interested in the speed with which Procedure. Internet users from Amazon Mechanical Turk they processed faces. In line with this cover story, participants saw 10 faces completed a study about their perceptions of other people. The with a yellow circle on them at random intervals throughout the adaptation study announcement made no mention of sex, gender, visual period. When one of these faces appeared, participants were told to press the enter key as quickly as possible. We did not record the speed with adaptation, or bias. After providing consent, participants evaluated 1 which participants identified the marked faces; we merely used this task to either male or female faces in several stages. First, participants ensure that participants remained engaged throughout the adaptation. We identified the most average-looking man or woman from an array used this strategy in all four studies reported here. FACE AFTEREFFECTS AND GENDER-RELATED BIASES 1263

Figure 1. Example stimuli depicting the range of masculine (top left) to feminine (top right) male faces as well as feminine (bottom left) to masculine (bottom right) female faces. The most gender-atypical face in each sex category (i.e., feminine male and masculine female) was the anthropomorphic mean at which the face had equally masculine and feminine characteristics. Because we informed participants of the sex category that they were judging, we refer to these faces as hyperfeminine and hypermasculine, although they were phenotypically androgynous.

multilevel regression models to account for the nested structure of ceptual norms for both men and women (ts ϭ 6.08 and 5.50, ps Ͻ the data. Specifically, we stacked Perceptual Gender Norm scores .001, respectively). in the data set and differentiated them with a dummy-coded Visual adaptation influences perceptual norms for male and variable (Test Period;0ϭ Pretest,1ϭ Posttest). The masculinity female faces. Next, we tested whether visual adaptation pro- and femininity dimensions of the PAQ were internally consistent duced an aftereffect that reliably altered perceptual norms for (Cronbach’s ␣s ϭ .81 and .84, respectively), so we summed the men’s and women’s faces. To do so, we regressed Perceptual items into continuous composite scores on which higher values Gender Norm onto Adaptation Condition, Test Period, Target Sex, indicated more masculine and feminine gender schemas. We and all interactions. The three-way interaction was not significant scored Participant Age (years) and Education (1 ϭ less than high (B ϭϪ1.04, SE ϭ 1.21, t ϭϪ0.86, p ϭ .390), indicating that the school to 8 ϭ doctoral degree) continuously. We effect-coded predicted aftereffects did not vary as a function of Target Sex. We Adaptation Condition, Target Sex, and Participant Sex (masculine therefore collapsed across Target Sex and regressed Perceptual adaptation ϭϪ0.5, feminine adaptation ϭ 0.5; male ϭϪ0.5, Gender Norm onto Adaptation Condition, Test Period, and their female ϭ 0.5), and dummy-coded Participant Race (White as interaction. The expected two-way interaction was significant reference category). (B ϭ 3.76, SE ϭ 0.60, t ϭ 6.23, p Ͻ .001). We decomposed this As noted above, we tested our hypotheses using random coef- interaction by examining simple slopes for Test Period within each ficient multilevel models to account for within-subject dependen- adaptation condition. In the masculine adaptation condition, par- cies in the data (i.e., multiple judgments of the same faces at ticipants chose significantly more masculine faces as normative at multiple time points). Although we included random intercepts to posttest relative to pretest (B ϭϪ2.50, SE ϭ 0.40, t ϭϪ6.15, p Ͻ account for the nested structure of the data, we were only inter- .001). In the feminine adaptation condition, the opposite trend ested in fixed effects; thus, we do not discuss random effects any emerged, such that participants chose significantly more feminine This document is copyrighted by the American Psychological Association or one of its allied publishers. further. Below, we report unstandardized regression coefficients, faces as normative at posttest relative to pretest (B ϭ 1.26, SE ϭ This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. which indicate effect size as the expected increase in the dependent 0.45, t ϭ 2.83, p ϭ .005) (see Figure 2). variable given a one-unit increase in the independent variable. We tested the strength of these effects by controlling for demo- Preliminary analyses. First, we examined the extremity of graphic variables collected during the study. We were particularly the gendered features that participants identified as normative at interested in whether gender schematicity altered adaptation ef- pretest. On the basis of anecdotal evidence that perceivers have fects, because the tendency to view the world in a starkly gendered greater exposure to feminine relative to highly masculine individ- manner might create strong perceptual gender norms that are uals, we expected that perceivers would enter the study with resistant to change. Although we did not have a priori hypotheses somewhat feminized norms for male and female faces. As ex- about the effects of other demographic variables on visual adap- pected, participants identified an anthropometrically feminized tation, we also controlled for them to assess the robustness of the male face (M ϭ 13.92, SD ϭ 4.96) and female face (M ϭ 13.39, observed aftereffects. Specifically, we recalculated the regressions SD ϭ 4.53) as normative at pretest. Both of these means were described above after partialing out the effects of participant sex, significantly above the midpoint of the gender scale (i.e., 11), age, race, education, and masculine and feminine gender schemas. indicating that participants entered the study with feminized per- After accounting for these factors, the two-way interaction be- 1264 LICK AND JOHNSON

Method Participants. One hundred eighty-eight Internet users from the United States (71 men, 116 women, one unreported) completed an online study. Participants were diverse in terms of age (M ϭ 34.05 years, SD ϭ 13.15 years) and racial identity (70% White, 10% Asian, 10% Black, 7% Latino, 3% biracial), and most reported a high level of education (83% attended college). Stimuli. Stimuli included a subset of the faces described in Study 1. For the evaluation portion of the study, we used five faces of each sex that were evenly spaced in terms of gender (a hypermasculine face, a moderately masculine face, a neutral face, a moderately feminine face, a hyperfeminine face). For the adaptation portion of the study, we used the five most feminine faces of each sex for the feminine adaptation condition and the five most masculine faces of each sex for the masculine adaptation condition. Procedure. Internet users from Amazon Mechanical Turk completed a study about their perceptions of other people. The study announcement made no mention of sex, gender, visual Figure 2. Change in perceptual gender norms as a function of visual adaptation, or bias. After providing consent, participants were adaptation in Study 1. Perceptual norms became more masculine following randomly assigned to evaluate either five male or five female faces adaptation to masculine faces, but more feminine following adaptation to in a random order across 12 items measured on 10-point semantic feminine faces. Error bars represent standard errors of the mean within each adaptation condition. differential scales (Pretest). The evaluations were modeled after N. H. Anderson’s (1968) study of the most potent adjectives used to describe others, and they included unattractive–attractive, tween Adaptation Condition and Test Period remained significant appropriate–inappropriate (reverse scored), improper–proper, ϭ ϭ ϭ and of similar magnitude as before (B 3.79, SE 0.64, t 5.95, respectable–indecent (reverse scored), acceptable–unacceptable Ͻ p .001). Moreover, the simple effect of Test Period remained (reverse scored), in poor taste–in good taste, sincere–insincere significant in both the masculine and feminine adaptation condi- (reverse scored), honest–dishonest (reverse scored), loyal–disloyal (re- ϭϪ ϭ ϭϪ tions (Bs 2.52 and 1.27, SEs 0.43 and 0.47, ts 5.91 and verse scored), intelligent–unintelligent (reverse scored), dependable–not Ͻ 2.69, ps .001). dependable (reverse scored), warm–cold (reverse scored). Previ- Overall, Study 1 provided support for our hypothesis that per- ous studies have shown that these items reliably capture evaluative ceptual norms for men’s and women’s faces are calibrated on the biases related to gendered facial features (Lick & Johnson, 2013). basis of recent visual exposure. Perceivers entered the study with After evaluating the target faces, participants were randomly notably feminized norms for men’s and women’s faces, buttressing assigned to adaptation conditions as described in Study 1. In recent evidence that Western observers have more exposure to particular, participants repeatedly viewed either the five most feminized faces than to highly masculine faces of either sex (see feminine faces (Feminine Adaptation) or the five most masculine Adams, 2011; E. Anderson, 2009; Coad, 2008; Swain, 2006). faces (Masculine Adaptation) from the sex category to which they Importantly, these norms shifted following experimental exposure were assigned for 3 s each for a total of 3 min. Following to highly gendered exemplars: Perceivers who adapted to mascu- adaptation, participants reevaluated the same target faces from line phenotypes chose more masculine faces as normative at post- pretest on the scales described above (Posttest). Finally, partici- test relative to pretest, whereas perceivers who adapted to feminine pants completed the PAQ (Spence et al., 1975) and reported their phenotypes chose more feminine faces as normative at posttest sex, age, race, and education before being debriefed. relative to pretest. Thus, observers’ perceptual norms efficiently In total, 93 participants evaluated male faces (46 masculine

This document is copyrighted by the American Psychological Association or one of its alliedadapted publishers. to gendered features in target faces. adaptation, 47 feminine adaptation) and 95 participants evaluated

This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. female faces (49 masculine adaptation, 46 feminine adaptation). A Study 2 chi-square test revealed that the number of participants did not ␹2 ϭ ϭ In Study 1, adaptation to gendered features reliably altered differ significantly across these four cells, (1) 0.84, p .772. perceptual norms for men’s and women’s faces. We now turn to Results and Discussion our second hypothesis—namely, that perceptual norms are associated with social evaluations of targets. We predicted that partic- Our primary aims in Study 2 were to determine whether eval- ipants would harbor preexisting biases against hypermasculine uative judgments of targets were associated with perceptual gender faces because they are perceived as less normative than feminine norms and to test whether visual adaptation altered these judg- faces. Furthermore, we predicted that these biases would shift ments. We predicted that perceivers would express notable biases following adaptation, such that exposure to masculine faces would against targets who deviated from perceptual gender norms. Thus, attenuate bias against hypermasculine targets but exposure to because norms were notably feminized in Study 1, we expected a feminine faces would exacerbate bias against hypermasculine tar- stronger bias against masculine relative to feminine targets at gets. pretest. More importantly, we predicted that these biases would FACE AFTEREFFECTS AND GENDER-RELATED BIASES 1265

shift following adaptation, such that visual exposure to masculine We decomposed the interaction by examining simple slopes for faces would lead to more favorable evaluations of masculine Gendered Features within each sex category. Feminine faces of targets but less favorable evaluations of feminine targets, whereas both sexes were evaluated more positively than masculine faces at visual exposure to feminine faces would lead to more favorable pretest, though this trend was more pronounced for male targets evaluations of feminine targets but less favorable evaluations of relative to female targets (Bs ϭ 3.85 and 1.58, SEs ϭ 0.43 and masculine targets. 0.42, ts ϭ 8.98 and 3.78, ps Ͻ .001, respectively). Thus, as in Because the target faces varied incrementally in gender typical- previous research, participants expressed notable biases against ity, we numerically coded each face from 1 (hypermasculine)to5 masculine faces of both sexes at pretest. (hyperfeminine) in order to quantify their gendered features. We Visual adaptation influences gender-related biases. Next, analyzed targets’ gendered appearances continuously based on this we tested whether visual adaptation produced an aftereffect that interval scale (hereafter, Gendered Features). We computed reliably altered gender-related biases in social evaluation. To do within-subject reliability for the evaluative items using the method so, we regressed Evaluations onto Gendered Features, Target Sex, described by Cranford et al. (2006), which indicates a scale’s Adaptation Condition, Test Period, and all interactions. The four- ability to capture changes in judgments across a range of stimuli. way interaction was not significant (B ϭϪ0.55, SE ϭ 1.75, t ϭ ϭ The items showed acceptable reliability (RC .82), so we summed Ϫ0.32, p ϭ .750), indicating that the predicted aftereffects did not them into continuous composite scores on which higher values vary as a function of Target Sex. We therefore collapsed across ϭ indicated more favorable evaluations at pretest (Min 12.00, Target Sex and regressed Evaluations onto Gendered Features, ϭ ϭ ϭ ϭ Max 120.00; M 80.43, SD 19.28) and posttest (Min Adaptation Condition, Test Period, and all interactions. The ex- ϭ ϭ ϭ 12.00, Max 120.00; M 79.21, SD 20.76). Because partic- pected three-way interaction was significant (B ϭ 3.92, SE ϭ 0.89, ipants evaluated the faces at two time points, we again stacked t ϭ 4.41, p Ͻ .001). evaluations in the data set and differentiated them with a dummy- We decomposed the three-way interaction by examining simple ϭ ϭ coded variable (Test Period;0 Pretest Evaluations,1 Posttest slopes of Gendered Appearance, Test Period, and their interaction Evaluations). As before, the masculinity and femininity dimen- within the masculine and feminine adaptation conditions (see ␣ϭ sions of the PAQ were internally consistent (Cronbach’s .81 Figure 3). In the masculine adaptation condition, the two-way and .86, respectively), so we created continuous composites for interaction between Gendered Appearance and Test Period was each subscale. We examined demographic variables exactly as significant (B ϭϪ1.70, SE ϭ 0.62, t ϭϪ2.72, p ϭ .007), described in Study 1, and we again tested our hypotheses using indicating that the effect of Gendered Appearance on Evaluations random coefficient multilevel models to account for within-subject differed significantly at pretest and posttest. At pretest, participants dependencies in the data. in the masculine adaptation condition evaluated feminine targets Gender-related bias at pretest. First, we tested whether per- more favorably than masculine targets (B ϭ 2.90, SE ϭ 0.44, t ϭ ceivers expressed biases against targets who deviated from per- 6.60, p Ͻ .001). At posttest, participants in the masculine adapta- ceptual norms. To do so, we created a Norm Deviation variable by tion condition showed less bias toward feminine targets; in fact, subtracting the average Perceptual Gender Norm for men and the magnitude of the bias was reduced by more than half (B ϭ women from Study 1 from each target’s objective gender score, 1.20, SE ϭ 0.44, t ϭ 2.71, p ϭ .007). based on the 21-point scale described in Study 1 (from 1 ϭ very masculine to 21 ϭ very feminine). We then took the absolute value In the feminine adaptation condition, the two-way interaction of this variable, such that norm deviation indicated absolute devi- between Test Period and Gendered Appearance was also signifi- ϭ ϭ ϭ Ͻ ation from the perceptual norm in either direction (hereafter, Norm cant (B 2.22, SE 0.63, t 3.51, p .001), indicating that the Deviation). For example, the average Perceived Gender Norm for effect of Gendered Appearance on Evaluations differed reliably at men in Study 1 was 13.92, and the most masculine male face had pretest and posttest. At pretest, participants in the feminine adap- a gender score of 1. So, the norm deviation score for this target was tation condition evaluated feminine targets more favorably than ϭ ϭ ϭ Ͻ 12.92, indicating a 12.92-point departure from observers’ percep- masculine targets (B 2.54, SE 0.45, t 5.68, p .001). At tual gender norms for men. Then we regressed Pretest Evaluations posttest, participants in the feminine adaptation condition were onto Norm Deviation. As expected, targets who deviated from the more biased, evaluating feminine targets even more favorably than perceived gender norm in either direction (i.e., more feminine or masculine targets; in fact, the magnitude of the bias nearly doubled This document is copyrighted by the American Psychological Association or one of its allied publishers. more masculine) were evaluated more negatively than targets who (B ϭ 4.76, SE ϭ 0.45, t ϭ 10.61, p Ͻ .001). This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. fell closer to the norm (B ϭϪ1.59, SE ϭ 0.15, t ϭϪ10.49, p Ͻ We tested the strength of these effects by controlling for demo- .001). graphic variables collected during the study. Specifically, we re- Because perceived norms were notably feminized in Study 1, calculated the regressions described above, partialing out the ef- there was more room for targets to deviate in the masculine fects of participant sex, age, race, education, and masculine and direction than in the feminine direction. This may help to explain feminine gender schemas. After accounting for these factors, fem- why several recent studies have uncovered evaluative biases inine male and female targets were still evaluated more favorably against hypermasculine targets—they simply deviate more from than masculine male and female targets at pretest (Bs ϭ 3.91 and the norm than do feminine targets. We tested whether perceivers 1.62, SEs ϭ 0.45 and 0.43, zs ϭ 8.69 and 3.76, ps Ͻ .001, expressed biases against hypermasculine targets in our study by respectively). Moreover, the three-way interaction indicating regressing Pretest Evaluations onto Gendered Features, Target change in evaluations from pretest to posttest as a function of Sex, and their interaction. The two-way interaction was significant Adaptation Condition and Gendered Appearance remained highly (B ϭϪ2.28, SE ϭ 0.60, t ϭϪ3.80, p Ͻ .001), indicating that significant and in the same direction as before (B ϭ 4.12, SE ϭ preferences for gendered features differed across sex categories. 0.92, t ϭ 4.47, p Ͻ .001). 1266 LICK AND JOHNSON

Figure 3. Change in evaluative judgments from pretest to posttest as a function of gendered features and adaptation condition in Study 2. The magnitude of the preference for feminine faces was smaller following adaptation to masculine features (left panel), but larger following adaptation to feminine features (right panel). Error bars depict standard errors for the effect of gendered features within each adaptation condition.

Overall, Study 2 provided support for our hypothesis that visual lamp, park bench, and chair). As in the other conditions, these adaptation alters gender-related biases in social perception. Per- images were repeatedly presented in random order for 3 s each for ceivers expressed notable biases against hypermasculine men and a total of 3 min. women after mere glimpses at their faces, in part because they In total, 192 participants evaluated male faces (63 masculine departed from perceptual gender norms. Exposure to a series of adaptation, 63 feminine adaptation, 66 control) and 155 partici- feminine faces exacerbated this bias, but exposure to masculine pants evaluated female faces (58 masculine adaptation, 47 femi- faces reduced it. These changes in evaluative judgments mirrored nine adaptation, 50 control). We collected data for male and the changes in perceptual gender norms that we observed in Study female targets for equal and prespecified amounts of time (2 1, suggesting that adaptation to gendered facial features altered weeks), but still obtained a larger sample for female targets. One evaluative preferences by making certain features appear more or reason for this difference in sample size may be that our studies less normative. In summary, visual exposure to gendered pheno- were quite demanding relative to other Mechanical Turk studies, types reliably altered gender-related biases in social evaluation. resulting in a limited pool of interested participants. Because we collected data for male targets during the 2 weeks immediately Study 3 preceding data collection for female targets, we suspect that we tapped the participant pool, resulting in a somewhat smaller sam- In Studies 1 and 2, adaptation to gendered facial features altered ple for female faces than for male faces. Importantly, however, a perceived phenotypic norms and subsequent social biases toward chi-square test revealed that the number of participants did not men and women. Although compelling, the lack of a control differ significantly across the six cells, ␹2(2) ϭ 0.81, p ϭ .669. condition did not allow us to specify whether visual adaptation shifted evaluations over and above any changes that may have Results and Discussion occurred due to repeated evaluations of the same stimuli (see Zajonc, 1968). Study 3 addressed this issue. Our primary aim in Study 3 was to replicate our previous findings with the addition of a control group. As before, we This document is copyrighted by the American Psychological Association or one of its allied publishers. expected to uncover a bias against masculine relative to feminine This article is intended solely for the personal use ofMethod the individual user and is not to be disseminated broadly. targets at pretest. Moreover, we predicted that these biases would Participants. Three hundred forty-seven Internet users from shift following adaptation, such that visual exposure to masculine the United States (151 men, 173 women, 23 unreported) completed faces would lead to more favorable evaluations of masculine an online study. Participants were diverse in terms of age (M ϭ targets but less favorable evaluations of feminine targets, whereas 33.12 years, SD ϭ 11.92 years) and racial identity (70% White, visual exposure to feminine faces would lead to more favorable 7% Asian, 8% Black, 9% Latino, 6% biracial), and most reported evaluations of feminine targets but less favorable evaluations of a high level of education (82% attended college). masculine targets. We did not expect the control condition to Stimuli and procedure. Stimuli and procedure were identical reliably alter gender-related biases in social perception. to Study 2, with the addition of a control condition. Participants in We examined our hypotheses in the same steps described in the control condition evaluated target faces exactly as described in Study 2. As before, we numerically coded each face from 1 Study 2, but during the adaptation phase, they viewed a series of (hypermasculine)to5(hyperfeminine) in order to quantify their images from the Psychological Image Collection at Stirling data- gendered appearance (hereafter, Gendered Features). Again, the base of manmade objects (www.pics.stir.ac.uk; trashcan, stapler, Evaluative Judgments items showed adequate within-subject reli- FACE AFTEREFFECTS AND GENDER-RELATED BIASES 1267

ϭ ability (RC .83), so we summed the items into continuous against hypermasculine targets in our study by regressing Pretest composite scores on which higher values indicated more favorable Evaluations onto Gendered Features, Target Sex, and their inter- evaluations at pretest (Min ϭ 0.00, Max ϭ 120.00; M ϭ 82.74, action. The two-way interaction was significant (B ϭϪ2.84, SE ϭ SD ϭ 19.99) and posttest (Min ϭ 0.00, Max ϭ 120.00; M ϭ 0.45, t ϭϪ6.34, p Ͻ .001), indicating that preferences for gen- 81.58, SD ϭ 21.43). Because participants evaluated the faces at dered features differed across sex categories. We decomposed this two time points, we stacked evaluation scores in the data set and interaction by examining simple slopes for Gendered Features differentiated them with a dummy-coded variable (Test Period; within each sex category. Feminine faces of both sexes were 0 ϭ Pretest Evaluations,1ϭ Posttest Evaluations). The mascu- evaluated more positively than masculine faces at pretest, though linity and femininity dimensions of the PAQ were internally con- this trend was more pronounced for male targets relative to female sistent (Cronbach’s ␣ϭ.78 and .84, respectively), so we created targets (Bs ϭ 4.07 and 1.23, SEs ϭ 0.31 and 0.33, zs ϭ 13.32 and continuous composite scores for each. All other numeric coding 3.77, ps Ͻ .001, respectively). Thus, replicating our effects from was identical to Study 2 with the exception of Adaptation Condi- Study 2, participants expressed notable biases against masculine tion, which was dummy coded (control as the reference category). faces of both sexes at pretest. As before, we tested our hypotheses using random coefficient Visual adaptation influences gender-related biases. Next, multilevel models to account for within-subject dependencies in we tested whether visual adaptation produced an aftereffect that the data. reliably altered gender-related biases in social evaluation. Because Gender-related bias at pretest. First, we tested whether per- Adaptation Condition was a multicategorical variable, we exam- ceivers expressed biases against targets who deviated from per- ined Type 3 tests of fixed effects to determine the significance of ceptual norms. As in Study 2, we created a norm deviation variable interactions. First, we regressed Evaluations onto Gendered Fea- by subtracting the average Perceptual Gender Norm for men and tures, Target Sex, Adaptation Condition, Test Period, and all women from Study 1 from each target’s gender score, based on the interactions. The four-way interaction was not statistically signif- 21-point scale described in Study 1 (from 1 ϭ very masculine to icant, ␹2(2) ϭ 1.25, p ϭ .287, indicating that the predicted after- 21 ϭ very feminine). We then took the absolute value of this effects did not vary as a function of Target Sex. We therefore variable, such that norm deviation indicated absolute deviation collapsed across Target Sex and regressed Evaluations onto Gen- from the perceptual norm in either direction. Then, we regressed dered Features, Adaptation Condition, Test Period, and all inter- Pretest Evaluations onto Norm Deviation. As expected, targets actions. The expected three-way interaction was significant, who deviated from the perceptual norm in either direction (i.e., ␹2(2) ϭ 4.83, p ϭ .008. more masculine or more feminine) were evaluated more negatively We decomposed the three-way interaction by examining simple than targets who fell closer to the norm (B ϭϪ1.60, SE ϭ 0.09, slopes of Gendered Features, Test Period, and their interaction t ϭϪ17.04, p Ͻ .001). within each adaptation condition (see Figure 4). In the masculine Because perceptual norms were notably feminized in Study 1, adaptation condition, the two-way interaction between Test Period there was more room for targets to deviate in the masculine and Gendered Features was marginally significant (B ϭϪ1.06, direction than in the feminine direction. This may help to explain SE ϭ 0.56, t ϭϪ1.88, p ϭ .060). At pretest, participants in the why previous studies have uncovered biases against hypermascu- masculine adaptation condition evaluated feminine targets more line targets—they simply differ more from the norm than do favorably than masculine targets (B ϭ 2.78, SE ϭ 0.39, t ϭ 7.07, feminine targets. We tested whether perceivers expressed biases p Ͻ .001). At posttest, participants in the masculine adaptation This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Figure 4. Change in evaluative judgments from pretest to posttest as a function of gendered features and adaptation condition in Study 3. The magnitude of the preference for feminine faces was smaller following adaptation to masculine faces (left panel), larger following adaptation to feminine faces (middle panel), and the same following a control adaptation (right panel). Error bars depict standard errors for the effect of gendered features within each adaptation condition. 1268 LICK AND JOHNSON

condition showed less bias toward feminine targets; in fact, the precise mechanism by which evaluative biases change following magnitude of the bias was reduced by a full scale point (B ϭ 1.72, visual adaptation to phenotypically natural facial cues. For now, SE ϭ 0.40, t ϭ 4.29, p Ͻ .001). we conclude that adaptation to both masculine and feminine faces In the feminine adaptation condition, the two-way interaction alters gender-related biases in social perception over and above between Test Period and Gendered Features was significant (B ϭ any effects of repeated evaluation. 1.46, SE ϭ 0.60, t ϭ 2.45, p ϭ .014). At pretest, participants in the feminine adaptation condition evaluated feminine targets more Study 4 favorably than masculine targets (B ϭ 3.04, SE ϭ 0.42, t ϭ 7.27, p Ͻ .001). At posttest, participants in the feminine adaptation In Studies 1–3, adaptation to gendered facial features affected condition were more biased, evaluating feminine targets even more both perceptual norms and social biases. Although this pattern of favorably than masculine targets; in fact, the magnitude of the bias results is consistent with existing theory, our conclusions are was more than a full scale point larger than it was to begin with limited by the fact that our stimuli relied on a single base face that (B ϭ 4.50, SE ϭ 0.42, t ϭ 10.61, p Ͻ .001). we morphed using computer software. Thus, it remains possible In the control condition, the two-way interaction between Test that the adaptation effects observed in our initial studies were due Period and Gendered Features was not significant (B ϭ 0.44, SE ϭ to idiosyncratic aspects of the base face (e.g., luminance, expres- 0.60, t ϭ 0.74, p ϭ .459). At pretest, participants in the control sion) or the gender-morphing program. Furthermore, it is worth condition evaluated feminine targets more favorably than mascu- noting that the stimuli in Studies 1–3 were bald, which may have line targets (B ϭ 2.40, SE ϭ 0.42, t ϭ 5.68, p Ͻ .001). At posttest, biased gendered perceptions of the faces toward masculinity be- participants in the control condition continued to evaluate feminine cause baldness is relatively rare among women. If anything, we targets more favorably than masculine targets, and this effect was suspect that this weakened the feminization of the perceptual similar in magnitude to the effect at pretest (B ϭ 2.84, SE ϭ 0.42, norms in Study 1 and the feminine adaptation effects in Studies 2 t ϭ 6.70, p Ͻ .001). Thus, the bias toward feminine targets did not and 3, both of which were highly significant. Nevertheless, repli- differ significantly following the control adaptation. cating our results with real faces of multiple individuals that vary We tested the strength of these effects by controlling for demo- naturally in gendered features and are cropped to remove hair cues graphic variables collected during the study. Specifically, we re- would improve the generalizability and applicability of our find- calculated the regressions described above, partialing out the ef- ings. fects of participant sex, age, race, education, and masculine and Furthermore, although our measure of social evaluations was feminine gender schemas. After accounting for these factors, fem- statistically reliable and based on previously published research inine male and female targets were still evaluated more favorably about interpersonal biases in face perception (Lick & Johnson, than masculine male and female targets at pretest (Bs ϭ 3.94 and 2013), some of the items may have tapped into participants’ beliefs 1.51, SEs ϭ 0.31 and 0.34, zs ϭ 12.74 and 4.44, ps Ͻ .001, about norms (e.g., appropriate–inappropriate) rather than evalua- respectively). Furthermore, the three-way interaction indicating tive preferences per se. Moreover, the items may have described change in evaluations from pretest to posttest as a function of traits that are more stereotypically desirable in women than in men Adaptation Condition and Target Gender remained highly signif- (e.g., proper, sincere, honest, warm), biasing our results toward icant and in the same direction as before, ␹2(2) ϭ 4.19, p ϭ .015. favorable evaluations of feminine targets. Replicating our effects Thus, Study 3 provided additional evidence for our hypothesis with more global measures of liking would ensure that our initial that visual adaptation helps to explain evaluative biases stemming findings were not due to the specific items included in the evalu- from gendered phenotypes. In particular, perceivers expressed ations scale. notable biases against hypermasculine men and women after mere Finally, we have argued that shifting perceptual norms are the glimpses at their faces, in part because masculine faces appeared mechanism by which visual adaptation alters social evaluations less normative than feminine faces. The addition of a control related to gendered phenotypes. Our initial findings were consis- condition allowed us to explore the directional nature of changes in tent with this possibility. Indeed, in Study 1, visual adaptation these biases following adaptation to highly gendered exemplars. shifted perceptual norms for men’s and women’s faces, such that For both male and female faces, adaptation to feminine features adaptation to hypermasculine features made masculine faces ap- exacerbated the bias, whereas adaptation to masculine features pear more normative, whereas adaptation to hyperfeminine fea- This document is copyrighted by the American Psychological Association or one of its allied publishers. reduced it. Importantly, the control condition had no significant tures made feminine faces appear more normative. Studies 2 and 3 This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. impact on evaluative biases against hypermasculine faces. showed that evaluative judgments shift in the same direction as We should note that the mechanism driving shifts in evaluative perceptual norms—masculine faces were evaluated more favor- biases after visual adaptation remains somewhat unclear. As ably following adaptation to hypermasculine features, whereas shown in Figure 4, following masculine adaptation, preferences for feminine faces were evaluated more favorably following adapta- masculine faces improved slightly, whereas preferences for femi- tion to hyperfeminine features. Although these findings suggest nine faces decreased. Following feminine adaptation, preferences that changes in perceptual norms may be functionally related to for feminine faces did not change appreciably, but preferences for shifts in evaluative biases related to gendered phenotypes, we masculine faces decreased. Thus, adaptation may affect social documented the effects in separate studies. Concurrently examin- biases by increasing preferences for targets similar to the adapting ing perceptual norms and evaluative biases in a single study would stimuli or by reducing preferences for targets dissimilar to the provide stronger evidence that shifting perceptual norms is the adapting stimuli. Our study provides some indication of both mechanism underlying the aftereffects we observed. trends, though the latter was especially strong. Future studies that With these considerations in mind, we designed Study 4 to also include control conditions will be critical to tease apart the replicate and extend our previous findings with a broader measure FACE AFTEREFFECTS AND GENDER-RELATED BIASES 1269

of social evaluation and real faces that varied naturally in their participants reevaluated the same target faces from pretest on the gendered phenotypes. Furthermore, we sought to systematically scales described above (Posttest). Finally, participants completed test shifting perceptual norms as a mechanism underlying these the PAQ (Spence et al., 1975) and reported their sex, age, race, and aftereffects. We predicted that biases against hypermasculine faces education before being debriefed. would shift following adaptation, such that exposure to masculine In total, 175 participants evaluated male faces (81 masculine faces would attenuate bias against hypermasculine targets but adaptation, 94 feminine adaptation) and 197 participants evaluated exposure to feminine faces would exacerbate bias against hyper- female faces (98 masculine adaptation, 99 feminine adaptation). A masculine targets, and that these changes would be associated with chi-square test revealed that the number of participants did not changes in perceived gender norms. differ across these four cells, ␹2(1) ϭ 2.67, p ϭ .10.

Method Results and Discussion Participants. Three hundred seventy-two Internet users from Our primary aim in Study 4 was to determine whether the the United States (160 men, 193 women, 19 unreported) completed adaptation effects we uncovered in Studies 1–3 generalized to real an online study. Participants were diverse in terms of age (M ϭ faces and to broader measures of social evaluation. We predicted 36.30 years, SD ϭ 13.82 years) and racial identity (76% White, that perceivers would express notable biases against masculine 7% Asian, 8% Black, 6% Latino, 2% biracial), and most reported targets at pretest and that these biases would shift following a high level of education (87% attended college). adaptation. Specifically, we expected that visual exposure to mas- Stimuli. Stimuli were 14 faces of real men and 14 faces of real culine faces would lead to more favorable evaluations of mascu- women that varied in their gendered features (100 ϫ 161 pixels at line targets but less favorable evaluations of feminine targets, 72 pixels/inch). Research assistants gathered these faces via Inter- whereas visual exposure to feminine faces would lead to more net searches for individuals who had extremely gendered appear- favorable evaluations of feminine targets but less favorable eval- ances (i.e., hypermasculine and hyperfeminine), were not celebri- uations of masculine targets. ties, and had no visible facial hair or tattoos. This resulted in a Because the target faces did not vary incrementally in gender sample of 88 images. Nineteen raters then evaluated the gendered typicality, we could not numerically code the faces as we did in our appearance of each image on a 10-point scale (1 ϭ Very Masculine previous studies. Instead, we analyzed targets’ gendered features to 10 ϭ Very Feminine). On the basis of these ratings, we selected on the basis of the ratings provided by independent coders (see the the seven most masculine and seven most feminine faces in each Stimuli section; 1 ϭ Very Masculine to 10 ϭ Very Feminine). sex category and then transformed the images to grayscale, stan- Specifically, we used the average gender score for each face as a dardized their size, and oval-cropped them to remove hair. A continuous measure of gendered appearance, on which higher separate sample of 12 raters evaluated the gendered appearance of values indicated a more feminine appearance (mean centered). The these standardized stimuli, rating the masculine male and female three interpersonal evaluations (attractiveness, likability, desire for ϭ ϭ faces (Ms 1.94 and 4.69, respectively) as notably more mascu- contact) showed high within-subject reliability (RC 0.93), so we line than the feminine male and female faces (Ms ϭ 6.33 and 9.47, summed them into continuous composite scores on which higher respectively) on a 10-point scale (from 1 ϭ Very Masculine to values indicated more favorable evaluations at pretest (Min ϭ 10 ϭ Very Feminine). Importantly, the raters were highly consis- 0.00, Max ϭ 30.00; M ϭ 14.25, SD ϭ 7.23) and posttest (Min ϭ tent in their judgments of the stimuli’s gendered appearances 0.00, Max ϭ 30.00; M ϭ 13.84, SD ϭ 7.73). We analyzed ϭ ϭ ϭ ϭ ϭ (ICCavg .97). perceptual norms at pretest (Min 0, Max 10; M 4.65, SD Procedure. Internet users from Amazon Mechanical Turk 2.52) and posttest (Min ϭ 0, Max ϭ 10; M ϭ 4.54, SD ϭ 2.52) on completed a study about their perceptions of other people. The the basis of participants’ ratings of the averageness of each face for study announcement made no mention of sex, gender, visual its sex category. Because participants evaluated the faces at two adaptation, or bias. After providing consent, participants were time points, we stacked these scores in the data set and differen- randomly assigned to rate either six of the male or six of the female tiated them with a dummy-coded variable (Test Period;0ϭ faces described above (three hypermasculine, three hyperfeminine) Pretest,1ϭ Posttest). As before, the masculinity and femininity across four items measured on 11-point sliding scales, with higher dimensions of the PAQ showed adequate reliability (Cronbach’s This document is copyrighted by the American Psychological Association or one of its allied publishers. values indicating more favorable ratings (Pretest). Three of these ␣ϭ.76 and .79, respectively), so we created continuous compos- This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. items assessed interpersonal judgments: (1) How attractive is this ite scores for each subscale. We examined demographic variables man(woman)? (2) How much do you like this man(woman)? (3) as described in Study 2, and again, we tested our hypotheses using How much would you like to be friends with this man(woman)? A random coefficient multilevel models to account for within-subject fourth item assessed perceptual gender norms: How much does dependencies in the data. this man(woman) look like the average man(woman)? Gender-related bias at pretest. First, we sought to replicate After evaluating the target faces, participants were randomly our finding that perceivers express notable biases against targets assigned to adaptation conditions as described in Studies 1 and 2. with highly masculine phenotypes. To do so, we regressed Pretest In particular, participants repeatedly viewed either the four remain- Evaluations onto Gendered Appearance, Target Sex, and their ing hyperfeminine faces (Feminine Adaptation) or the four remain- interaction. The two-way interaction was significant (B ϭ 1.61, ing hypermasculine faces (Masculine Adaptation) from the sex SE ϭ 0.09, t ϭ 17.29, p Ͻ .001), indicating that preferences for category to which they were assigned for 3 s each for a total of 3 gendered features differed across sex categories. We decomposed min. Thus, in contrast to our previous studies, the adaptation the interaction by examining simple slopes for Gendered Appear- stimuli differed from the evaluation stimuli. Following adaptation, ance within each sex category. As before, feminine faces of both 1270 LICK AND JOHNSON

sexes were evaluated more favorably than masculine faces at We tested the strength of these effects by controlling for demo- pretest, though this trend was more pronounced for female targets graphic variables collected during the study. Specifically, we re- relative to male targets (Bs ϭ 0.41 and 2.02, SEs ϭ 0.07 and 0.06, calculated the regressions described above, partialing out the ef- ts ϭ 5.57 and 36.16, ps Ͻ .001, respectively). These findings differ fects of participant age, sex, race, education, and masculine and slightly from Study 2, in which the bias against masculine faces feminine gender schemas. After accounting for these factors, fem- was stronger for men than women. Despite this discrepancy, the inine male and female targets were still evaluated more favorably effect was highly significant for both sexes, indicating a strong than masculine male and female targets at pretest (Bs ϭ 0.41 and bias against hypermasculine faces at pretest. 1.99, SEs ϭ 0.08 and 0.06, ts ϭ 5.23 and 33.20, ps Ͻ .001, Visual adaptation influences gender-related biases. Next, respectively). Moreover, the three-way interaction indicating we tested whether visual adaptation produced an aftereffect that change in evaluations from pretest to posttest as a function of reliably altered gender-related biases in social evaluation. To do Adaptation Condition and Target Gender remained marginally so, we regressed Evaluations onto Gendered Appearance, Target significant and in the same direction as before (B ϭϪ0.22, SE ϭ Sex, Adaptation Condition, Test Period, and all interactions. The 0.11, t ϭϪ1.92, p ϭ .056). ϭϪ four-way interaction was not statistically significant (B 0.10, Perceptual norms drive observed shifts in evaluative biases. SE ϭ 0.27, t ϭϪ0.37, p ϭ .714), indicating that the predicted A final goal of Study 3 was to test whether changes in perceptual aftereffects did not vary as a function of Target Sex. We therefore norms drove the observed changes in evaluation following adap- collapsed across Target Sex and regressed Evaluations onto Gen- tation. We first tested the association between perceived norms and dered Appearance, Adaptation Condition, Test Period, and all evaluations by regressing Pretest Evaluations onto Pretest Percep- interactions. The expected three-way interaction was significant tual Norms. As expected, targets who appeared more normative at (B ϭϪ0.24, SE ϭ 0.11, t ϭϪ2.19, p ϭ .029) (see Figure 5). pretest were evaluated more favorably than were targets who We decomposed the three-way interaction by examining simple appeared less normative (B ϭ 0.85, SE ϭ 0.06, t ϭ 15.31, p Ͻ slopes of Gendered Appearance, Test Period, and their interaction within the masculine and feminine adaptation conditions. In the .001). The same trend emerged when we regressed Posttest Eval- masculine adaptation condition, the two-way interaction between uations onto Posttest Perceptual Norms: Targets who appeared Gendered Appearance and Test Period was not significant (B ϭ more normative at posttest were evaluated more favorably than ϭ ϭ Ϫ0.02, SE ϭ 0.08, t ϭϪ0.22, p ϭ .828). In the feminine were targets who appeared less normative (B 1.16, SE 0.06, ϭ Ͻ adaptation condition, however, the two-way interaction between t 19.46, p .001). Next, we restructured the data set and created Test Period and Gendered Appearance was significant (B ϭ 0.22, a change in evaluations variable (Posttest Evaluations – Pretest SE ϭ 0.08, t ϭ 2.81, p ϭ .005), indicating that the effect of Evaluations) and a change in perceptual corm variable (Posttest Gendered Appearance on Evaluations differed reliably at pretest Norm – Pretest Norm). Then, we regressed Change in Evaluations and posttest. At pretest, participants in the feminine adaptation onto Change in Perceptual Norm. As faces were perceived as condition evaluated masculine targets less favorably than feminine increasingly normative from pretest to posttest, they were also targets (B ϭ 1.48, SE ϭ 0.06, t ϭ 24.25, p Ͻ .001). At posttest, evaluated more favorably (B ϭ 0.39, SE ϭ .04, t ϭ 10.29, p Ͻ participants in the feminine adaptation condition were even more .001) (see Figure 6). Thus, perceptual norms and social evaluations biased, evaluating masculine targets less favorably than they did at changed concurrently, such that faces that appeared more norma- pretest (B ϭ 1.70, SE ϭ 0.06, t ϭ 27.00, p Ͻ .001). tive were also evaluated more favorably following adaptation. This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Figure 5. Change in evaluative judgments from pretest to posttest as a function of gendered features and adaptation condition in Study 4. The magnitude of the preference for feminine faces did not change following adaptation to masculine features (left panel), but was larger following adaptation to feminine features (right panel). Error bars depict standard errors for the effect of gendered features within each adaptation condition. FACE AFTEREFFECTS AND GENDER-RELATED BIASES 1271

faces, improving the chances of obtaining a feminine adaptation effect. Indeed, at least for the female stimuli, feminine faces were rated as being extremely feminine (9.47 on a 10-point scale ranging from masculine to feminine), whereas the masculine faces were only slightly below the midpoint of the scale (4.69 on a 10-point scale ranging from masculine to feminine). If feminine features were indeed more extreme than masculine features in our stimuli, then the feminine adaptation effect may have been stronger than the masculine adaptation effect, as we saw here. Despite the nonsignificant masculine adaptation effect, perceptual gender norms did help to explain changes in evaluative biases following feminine adaptation. In particular, we found that changes in perceptual norms predicted changes in evaluative biases related to gendered facial features. Accounting for these norm shifts greatly improved the fit of regression models predicting gender-related biases before and after visual adaptation to gendered facial features. It is also important to note that the magnitude of the feminine adaptation effect was smaller than in our previous studies. This smaller effect size offers an important caveat to our findings, Figure 6. Change in evaluative judgments from pretest to posttest as a suggesting that real-world visual aftereffects may be somewhat function of change in perceived gender norms in Study 4. Targets were limited when they occur after such brief exposure. In general, evaluated more favorably as they were perceived to be increasingly nor- however, Study 4 provided evidence that adaptation effects can mative. Error bars depict standard errors for the effect of perceived gender indeed emerge on broad measures of evaluative preference for real norms. faces that vary naturally in their gendered phenotypes, gesturing toward the generalizability and applicability of our findings.

As a final test of whether changes in perceptual norms helped to General Discussion explain the observed changes in evaluations following adaptation, Perceivers express strong biases against hypermasculine indi- we constructed a series of nested regression models. First, we viduals of both sexes after mere glimpses at their faces (Jones & regressed Evaluations onto Gendered Appearance, Test Period, Hill, 1993; Little & Hancock, 2002; Perrett et al., 1998; Rhodes et Adaptation Condition, and all interactions (Deviance ϭ 27222.90). al., 2000). Across four studies, we found that these biases were Next, we added Perceptual Norms and all interactions to the model associated with recent visual exposure to gendered facial features. (Deviance ϭ 26221.30). We then performed a likelihood ratio test Study 1 revealed that perceptual norms for men’s and women’s on the deviance values from these two models, which assesses faces shift depending on visual experience. In particular, exposure whether the inclusion of the new variable improved model fit. As to hypermasculine phenotypes made masculine faces of both sexes expected, the inclusion of Perceptual Norms significantly im- appear more normative, whereas exposure to hyperfeminine phe- proved the fit of the regression model, ␹2(7) ϭ 1001.60, p Ͻ .001. notypes made feminine faces of both sexes appear more normative. Thus, accounting for shifts in perceptual norms greatly improved These shifts in perceptual norms were mirrored by shifts in eval- our ability to predict evaluations on the basis of gendered pheno- uative judgments related to facial appearance: In Studies 2–4, brief types following visual adaptation. adaptation to hypermasculine phenotypes reduced biases against In summary, Study 4 partially replicated our previous findings masculine targets, but exposure to hyperfeminine phenotypes ex- by demonstrating that perceivers expressed notable biases against acerbated them. With one exception (i.e., a nonsignificant effect hypermasculine men and women after mere glimpses at their faces of masculine adaptation in Study 4), this pattern of effects and that perceivers who were experimentally exposed to feminine This document is copyrighted by the American Psychological Association or one of its allied publishers. replicated across both male and female targets, multiple out- faces developed even stronger biases against masculine faces. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. come measures, and real and computer-generated faces. More- Unexpectedly, adaptation to masculine faces did not reliably alter over, the findings were robust after controlling for a host of social evaluations related to gendered facial features. Although it potentially confounding factors (e.g., gender schemas). There- is always difficult to interpret null findings, the nonsignificant fore, visual adaptation affects perceptual norms and social masculine adaptation effect may have emerged for several reasons. evaluations related to gendered phenotypes. These results pro- First, the stimuli in Study 4 were less carefully controlled than the vide novel information about the perceptual mechanics of computer-generated stimuli in Studies 1–3. Thus, the masculine gender-related biases in particular, and they pinpoint important faces may have varied in unexpected ways (e.g., viewpoint) that new directions for research on impression formation and bias the feminine faces did not. Although adaptation effects can transfer reduction more generally. across changes in viewpoint and other low-level features (see Jeffery, Rhodes, & Busey, 2006), they may be substantially weak- Perceptual Underpinnings of Prejudice ened, reducing our chances of obtaining a significant effect for masculine adaptation in Study 4. It is also possible that features in The primary contribution of our work is a new framework for the feminine faces were more extreme than features in the masculine understanding the perceptual underpinnings of prejudice. Interper- 1272 LICK AND JOHNSON

sonal prejudice is an enduring psychological problem with insid- in the population. As such, the current work helps to extend the ious consequences for target groups (Dovidio & Gaertner, 2010; ecological validity of research on face aftereffects. Dovidio, Glick, & Rudman, 2005; Lick, Durso, & Johnson, in Furthermore, most previous studies of face aftereffects have press). One reason that prejudice may persist is its foundation in examined how habituation to facial features affects ratings of basic processes that guide human perception. Indeed, a growing subsequent targets’ normality and attractiveness, often with single- body of research suggests that people form impressions of others item measures. Our studies are among the first to demonstrate that on the basis of brief glimpses of their faces (Freeman et al., 2010; aftereffects alter a broader constellation of social judgments that Ito et al., 2004; Lick & Johnson, 2013) and that these impressions arise together early in person perception. In particular, we found have important implications for downstream behaviors (Snyder & that adaptation to gendered facial features affects perceivers’ judg- Stukas, 1999). Although researchers have begun to document these ments of many characteristics, including a person’s warmth, re- early footprints of prejudice, the perceptual processes that give rise spectability, sincerity, honesty, intelligence, and overall likability. to them remain less clear. Thus, our studies extend face aftereffects to a decidedly interper- Perceptual norms have emerged as one important mechanism sonal domain, revealing that visual adaptation alters a broad set of underlying interpersonal bias. Researchers have long known that social evaluations that have implications for impression formation perceivers tend to prefer prototypical exemplars to more unique and downstream social interactions. exemplars of a given category (Posner & Keele, 1968). More recent work has extended this concept to social domains. For Prejudice and Gendered Appearances example, one study revealed that targets whose facial features were normative for their sex category were likely to be categorized as Our findings also engage with long-standing discussions about heterosexual and evaluated favorably, whereas those whose phe- the effects of gendered portrayals in everyday life. Feminist schol- notypes were less normative for their sex category were likely to ars have argued that people strategically exaggerate gendered be categorized as gay and evaluated negatively (Lick & Johnson, aspects of their facial appearance (Davis, 2002; Gill, Henwood, & 2013). These findings generally suggest that perceptual norms help McLean, 2005; Schilt & Westbrook, 2009) and that exposure to to explain biases that originate in the early stages of person such highly gendered portrayals is damaging to both women and perception. Still, the factors that influence perceptual norms them- men (Prentice & Carranza, 2002). However, the specific mecha- selves have remained unclear. The current research begins to nisms linking gendered imagery to downstream social biases have reveal how perceptual norms are calibrated, and how these cali- remained largely unexplored. The current work reveals that expo- brations affect social evaluations. sure to highly gendered imagery alters low-level social cognitive Specifically, our findings suggest that perceptual norms for a processes. In particular, we found that even relatively brief expo- given social category change on the basis of a perceiver’s recent sure to gendered facial features can shift perceptual norms for men visual experiences with members of that category. In the case of and women and subsequently alter social judgments. Thus, our studies provide empirical evidence for previous claims that the gender, perceivers who encounter highly feminine phenotypes tend overrepresentation of highly gendered phenotypes perpetuates in- to form feminized facial norms, whereas those who encounter terpersonal biases. masculine phenotypes tend to form masculinized facial norms. Our findings also supplement previous research that has Furthermore, because perceivers tend to prefer targets who appear noted the frequency and consistency of gender-related biases in prototypical, these shifts in perceptual norms help to explain social perception (Little & Perrett, 2011; Perrett et al., 1998; evaluative biases against individuals who embody nonprototypical Rhodes et al., 2000; Rudman & Glick, 2001; Rudman, Green- features. Indeed, the current studies revealed that visual adaptation wald, & McGhee, 2001). Indeed, we found that perceivers judge to highly gendered phenotypes reliably shifted perceptual norms, hypermasculine men and women negatively across a wide range which in turn shifted evaluative judgments. of social domains. Furthermore, we pinpointed deviation from The current research therefore augments our knowledge of how perceptual gender norms as a factor contributing to these biases. visual exposure molds perceptual norms, with implications for Because the norms for men’s and women’s faces were initially downstream social evaluations. As such, our work uniquely con- feminized, faces could deviate more in the masculine direction tributes to psychologists’ understanding of the perceptual roots of than in the feminine direction. This may account for perceivers’ This document is copyrighted by the American Psychological Association or one of its alliedinterpersonal publishers. prejudice. Although our findings cannot speak to the negative evaluations of masculine targets relative to feminine This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. factors underlying development of perceptual norms in the first targets—they simply deviate further from perceptual norms. In place, they suggest that visual exposure reliably alters these norms, this way, our findings help to explain the somewhat surprising with concomitant effects on evaluative preferences. Our findings results of recent studies that have indicated preferences for also inform literatures related to visual aftereffects and prejudice feminized facial features in both sexes. reduction, which we discuss below. Prejudice Reduction Visual Aftereffects and Social Evaluations Finally, and perhaps most importantly, the current work has The current work contributes to the burgeoning literature on important implications for theory and applied efforts aimed at visual aftereffects. To date, most evidence of higher level face prejudice reduction. Classic studies demonstrated that repeated aftereffects has been limited to grossly distorted features (e.g., exposure to a stimulus increases liking for that stimulus (mere caricatures). Our studies are among a small handful demonstrating exposure; Zajonc, 1968, 2001), and more recent work extended that humans also adapt to gendered phenotypes that vary naturally these effects to social stimuli (Dasgupta & Greenwald, 2001; FACE AFTEREFFECTS AND GENDER-RELATED BIASES 1273

Mutz & Goldman, 2010; Schiappa et al., 2005). Despite a tation effect was relatively modest in Study 4, where we used real long-standing theoretical focus on mere exposure as a method faces. Before planning large-scale interventions, it will be impor- of prejudice reduction, the perceptual mechanisms underlying tant for researchers to determine the efficacy of relatively brief mere exposure have remained elusive. The current studies in- visual exposure for reducing prejudice. dicate that shifts in perceptual norms may underlie these well- documented effects—because repeated exposure to a particular type of stimulus renders it normative, repeated exposure en- Conclusion hances liking. In summary, people form impressions of others based on the Moreover, mere exposure effects have generally shown that perceived normativity of their gendered facial features; recent exposure to a stimulus increases liking for that same stimulus. visual exposure helps to determine these perceptual norms. Thus, Although some scholars have argued that exposure might also features with which perceivers have relatively limited exposure improve liking for similar but nonidentical targets, evidence of (e.g., hypermasculine features) appear nonnormative and tend to a generalized mere exposure effect is scant (Rhodes, Halber- be evaluated negatively. Although the current findings demon- stadt, Jeffery, & Palermo, 2005). Here, we found that visual strate the existence and perceptual underpinnings of gender-related exposure may indeed operate at the level of features. Viewing biases in social evaluation, they also offer the optimistic conclu- a series of highly gendered facial features enhanced perceivers’ sion that manipulating the composition of one’s visual environ- evaluations of individuals who share similar features, even ment can change biases. This information may aid in eradicating when those individuals were not included among the adaptation the unequal treatment of men and women on the basis of their stimuli (e.g., Study 4). These findings pave the way for con- gendered phenotypes, and may eventually lead us toward a better tinued research on generalized mere exposure effects, and they understanding of the formation and reduction of other biases suggest that visual adaptation may be an important process related to physical appearance. driving such effects. Proponents of contact theory have argued that repeated interpersonal encounters reduce prejudice against a stigmatized group References (Dovidio, Eller, & Hewstone, 2011; Pettigrew & Tropp, 2006). Adams, A. (2011). “Josh wears pink cleats:” Inclusive masculinity on the Whereas classical theorizing suggested that face-to-face interac- soccer field. Journal of Homosexuality, 58, 579–596. doi:10.1080/ tions between individuals of equal status working to achieve a 00918369.2011.563654 common goal were necessary to achieve the social benefits of Addams, R. (1834). An account of a peculiar optical phenomenon seen contact (Allport, 1954), newer studies have demonstrated that after having looked at a moving body. London and Edinburgh Magazine merely observing or imagining outgroup members can effectively and Journal of Science, 5, 373–374. reduce prejudice (Birtel & Crisp, 2012; Dasgupta & Greenwald, Allport, G. W. (1954). The nature of prejudice. New York, NY: Addison- 200l; Gómez, Tropp, & Fernández, 2011; Schiappa et al., 2005; Wesley. Wright, Aron, McLaughlin-Volpe, & Ropp, 1997). The current Anderson, E. (2005). Orthodox and inclusive masculinity: Competing masculinities among heterosexual men in a feminized terrain. Sociolog- studies implicate visual adaptation as a proximal mechanism un- ical Perspectives, 48, 337–355. doi:10.1525/sop.2005.48.3.337 derlying the benefits of observed contact. In particular, our find- Anderson, E. (2008). Inclusive masculinity in a fraternal setting. Men and ings suggest that perceivers visually adapt following repeated Masculinities, 10, 604–620. doi:10.1177/1097184X06291907 exposure to members of different social groups, which makes them Anderson, E. (2009). Inclusive masculinity: The changing nature of mas- appear more normative and subsequently leads to more favorable culunities. New York, NY: Routledge. social evaluations. Although higher level psychological processes Anderson, N. D., & Wilson, H. R. (2005). The nature of synthetic face may also contribute to contact effects (e.g., reduced intergroup adaptation. Vision Research, 45, 1815–1828. doi:10.1016/j.visres.2005 anxiety; Pettigrew, 1998), visual adaptation may act as a low-level .01.012 perceptual mechanism, which helps to explain why recent vicari- Anderson, N. H. (1968). Likeableness ratings of 555 personality-trait ous and extended contact can effectively reduce prejudice. In all, words. Journal of Personality and Social Psychology, 9, 272–279. doi:10.1037/h0025907 our findings address recent calls for a deeper understanding of the Anzures, G., Mondloch, C. J., & Lackner, C. (2009). Face adaptation and proximal mechanisms underlying several common methods of This document is copyrighted by the American Psychological Association or one of its allied publishers. attractiveness aftereffects in 8-year-olds and adults. Child Development, prejudice reduction (Dovidio et al., 2011).

This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. 80, 178–191. doi:10.1111/j.1467-8624.2008.01253.x Finally, in terms of practical application, our results offer some Bestelmeyer, P. E. G., Jones, B. C., DeBruine, L. M., Little, A. C., Perrett, preliminary evidence that media interventions may reduce inter- D. I., Schneider, A.,...Conway, C. A. (2008). Sex-contingent face personal biases that originate in the early moments of face per- aftereffects depend on perceptual category rather than structural encod- ception. Indeed, some social groups are underrepresented in na- ing. Cognition, 107, 353–365. doi:10.1016/j.cognition.2007.07.018 tional media (Williams, 2000), and individuals from these Birtel, M. D., & Crisp, R. J. (2012). “Treating” prejudice: An exposure- underrepresented groups also face consistent prejudice and dis- therapy approach to reducing negative reactions toward stigmatized crimination (Dovidio, Gaertner, Kawakami, & Hodson, 2002). Our groups. Psychological Science, 23, 1379–1386. doi:10.1177/ 0956797612443838 results suggest that deliberately exposing people to more visual Blanz, V., & Vetter, T. (1999). A morphable model for the synthesis of 3D representations of these underrepresented groups in film, televi- faces. In Proceedings of the 26th Annual SIGGRAPH Meeting (pp. sion, and print might help to mitigate biases that develop in the 187–194). Los Angeles, CA: Addison Wesley Longman. early stages of person perception and form the basis of down- Buckingham, G., DeBruine, L. M., Little, A. C., Welling, L. M., Conway, stream prejudice. Still, it is important to note that the masculine C. A., Tiddeman, B. P., & Jones, B. C. (2006). Visual adaptation to adaptation effect was not significant and that the feminine adap- masculine and feminine faces influences generalized preferences and 1274 LICK AND JOHNSON

perceptions of trustworthiness. Evolution and Human Behavior, 27, Harris, J., & Clayton, B. (2007). The first metrosexual rugby star: Rugby 381–389. doi:10.1016/j.evolhumbehav.2006.03.001 union, masculinity, and celebrity in contemporary Wales. Sociology of Carpinella, C. M., & Johnson, K. L. (2013). Appearance-based politics: Sport Journal, 24, 145–164. Sex-typed facial cues communicate political party affiliation. Journal of Heldman, C., & Wade, L. (2011). Sexualizing Sarah Palin: The social and Experimental Social Psychology, 49, 156–160. doi:10.1016/j.jesp.2012 political context of the sexual objectification of female candidates. Sex .08.009 Roles, 65, 156–164. doi:10.1007/s11199-011-9984-6 Carter, C., & Steiner, L. (2004). Critical readings: Media and gender. Hering, E. (1878). Zur Lehre vom Lichtsinne [Outlines of a theory of the London, England: Open University Press. light sense]. Vienna, Austria: Gerolds Sohn. Clifford, C. W. G., & Rhodes, G. (2005). Fitting the mind to the world: Ito, T. A., & Cacioppo, J. T. (2000). Electrophysiological evidence of Adaptation and after-effects in high-level vision. Oxford, England: Ox- implicit and explicit categorization processes. Journal of Experimental ford University Press. doi:10.1093/acprof:oso/9780198529699.001.0001 Social Psychology, 36, 660–676. Coad, D. (2008). The metrosexual. New York, NY: SUNY. Ito, T. A., Thompson, E., & Cacioppo, J. T. (2004). Tracking the time- Collins, R. L. (2011). Content analysis of gender roles in media: Where are course of social perception: The effects of racial cues on event-related we now and where should we go? Sex Roles, 64, 290–298. doi:10.1007/ brain potentials. Personality and Social Psychology Bulletin, 30, 1267– s11199-010-9929-5 1280. doi:10.1177/0146167204264335 Cranford, J. A., Shrout, P. E., Iida, M., Rafaeli, E., Yip, T., & Bolger, N. Jeffery, L., Rhodes, G., & Busey, T. (2006). View-specific coding of face (2006). A procedure for evaluating sensitivity to within-person change: shape. Psychological Science, 17, 501–505. doi:10.1111/j.1467-9280 Can mood measures in diary studies detect change reliably? Personality .2006.01735.x and Social Psychology Bulletin, 32, 917–929. doi:10.1177/ Johnson, K. L., Freeman, J. B., & Pauker, K. (2012). Race is gendered: 0146167206287721 How covarying phenotypes and stereotypes bias sex categorization. Dasgupta, N., & Greenwald, A. G. (2001). On the malleability of automatic Journal of Personality and Social Psychology, 102, 116–131. doi: attitudes: Combating automatic prejudice with images of admired and 10.1037/a0025335 disliked individuals. Journal of Personality and Social Psychology, 81, Johnston, V. S., & Franklin, M. (1993). Is beauty in the eye of the 800–814. doi:10.1037/0022-3514.81.5.800 beholder? Ethology and Sociobiology, 14, 183–199. doi:10.1016/0162- Davis, K. (2002). A dubious equality: Men, women and cosmetic surgery. 3095(93)90005-3 Johnston, V. S., Hagel, R., Franklin, M., Fink, B., & Grammer, K. (2001). Body & Society, 8, 49–65. doi:10.1177/1357034X02008001003 Male facial attractiveness: Evidence for hormone-mediated adaptive Dovidio, J. F., Eller, A., & Hewstone, M. (2011). Improving intergroup design. Evolution and Human Behavior, 22, 251–267. doi:10.1016/ relations through direct, extended and other forms of indirect contact. S1090-5138(01)00066-6 Group Processes & Intergroup Relations, 14, 147–160. doi:10.1177/ Jones, D., & Hill, K. (1993). Criteria of facial attractiveness in five 1368430210390555 populations. Human Nature, 4, 271–296. doi:10.1007/BF02692202 Dovidio, J. F., & Gaertner, S. L. (2010). Intergroup bias. In S. T. Fiske, Kahneman, D., & Miller, D. T. (1986). Norm theory: Comparing reality to D. T. Gilbert, & G. Lindzey (Eds.), Handbook of social psychology (pp. its alternatives. Psychological Review, 93, 136–153. doi:10.1037/0033- 1084–1121). Hoboken, NJ: Wiley. 295X.93.2.136 Dovidio, J. F., Gaertner, S. L., Kawakami, K., & Hodson, G. (2002). Why Keating, C. F. (1985). Gender and the physiognomy of dominance and can’t we just get along? Interpersonal biases and interracial distrust. attractiveness. Social Psychology Quarterly, 48, 61–70. doi:10.2307/ Cultural Diversity and Ethnic Minority Psychology, 8, 88–102. doi: 3033782 10.1037/1099-9809.8.2.88 Leopold, D. A., & Bondar, I. (2005). Adaptation to complex visual patterns Dovidio, J. F., Glick, P., & Rudman, L. (Eds.). (2005). On the nature of in humans and monkeys. In C. W. G. Clifford & G. Rhodes (Eds.), prejudice: Fifty years after Allport. Malden, MA: Blackwell. doi: Fitting the mind to the world: Adaptation and after-effects in high-level 10.1002/9780470773963 vision (pp. 189–212). Oxford, England: Oxford University Press. doi: Fang, F., Ijichi, K., & He, S. (2007). Transfer of the face viewpoint 10.1093/acprof:oso/9780198529699.003.0008 aftereffect from adaptation to different and inverted faces. Journal of Leopold, D. A., Rhodes, G., Muller, K. M., & Jeffery, L. (2005). The Vision, 7, 1–9. doi:10.1167/7.13.6 dynamics of visual adaptation to faces. Proceedings of the Royal Society Frederick, D. A., & Haselton, M. G. (2007). Why is muscularity sexy? of London, Series B, 272, 897–904. doi:10.1098/rspb.2004.3022 Tests of the fitness indicator hypothesis. Personality and Social Psy- Lick, D. J., Durso, L. E., & Johnson, K. L. (in press).Minority stress and chology Bulletin, 33, 1167–1183. doi:10.1177/0146167207303022 physical health in sexual minority communities. Perspectives on Psy- Freeman, J. B., Johnson, K. L., Ambady, N., & Rule, N. O. (2010). Sexual chological Science. This document is copyrighted by the American Psychological Association or one of its allied publishers. orientation perception involves gendered facial cues. Personality Lick, D. J., & Johnson, K. L. (2013). Fluency of visual processing explains This article is intended solely for the personal use of the individual userand and is not to be disseminated broadly. Social Psychology Bulletin, 36, 1318–1331. doi:10.1177/ prejudiced evaluations of minority targets. Journal of Experimental 0146167210378755 Social Psychology, 49, 419–425. doi:10.1016/j.jesp.2013.01.001 Gill, R., Henwood, K., & McLean, C. (2005). Body projects and the Little, A. C., & Hancock, P. J. B. (2002). The role of masculinity and regulation of normative masculinity. Body & Society, 11, 37–62. doi: distinctiveness in judgments of human male facial attractiveness. 10.1177/1357034X05049849 British Journal of Psychology, 93, 451–464. doi:10.1348/ Gómez, A., Tropp, L. R., & Fernández, S. (2011). When extended contact 000712602761381349 opens the door to future contact: Testing the effects of extended contact Little, A. C., Jones, B. C., Penton-Voak, I. S., Burt, D. M., & Perrett, D. I. on attitudes and intergroup expectancies in majority and minority (2002). Partnership status and the temporal context of relationships groups. Group Processes & Intergroup Relations, 14, 161–173. doi: influence human female preferences for sexual dimorphism in male face 10.1177/1368430210391119 shape. Proceedings of the Royal Society B, 269, 1095–1100. doi: Grammer, K., & Thronhill, R. (1994). Human (Homo sapiens) facial 10.1098/rspb.2002.1984 attractiveness and sexual selection: The role of symmetry and average- Little, A. C., & Perrett, D. I. (2011). Facial attractiveness. In R. B. Adams, ness. Journal of Comparative Psychology, 108, 233–242. doi:10.1037/ Jr., N. Ambady, & K. Nakayama (Eds.), The science of social vision (pp. 0735-7036.108.3.233 164–185). New York, NY: Oxford University Press. FACE AFTEREFFECTS AND GENDER-RELATED BIASES 1275

Manago, A. M., Graham, M. B., Greenfield, P. M., & Salimkhan, G. implicate norm-based coding of faces. In C. W. G. Clifford & G. Rhodes (2008). Self-presentation and gender on MySpace. Journal of Applied (Eds.), Fitting the mind to the world: Adaptation and after-effects in Developmental Psychology, 29, 446–458. doi:10.1016/j.appdev.2008 high-level vision (pp. 213–240). Oxford, England: Oxford University .07.001 Press. doi:10.1093/acprof:oso/9780198529699.003.0009 Mutz, D. C., & Goldman, S. K. (2010). Mass media. In J. F. Dovidio, M. Rudman, L. A., & Glick, P. (2001). Prescriptive gender stereotypes and Hewstone, P. Glick, & V. M. Esses (Eds.), Handbook of prejudice backlash toward agentic women. Journal of Social Issues, 57, 743–762. stereotyping, and discrimination (pp. 241–258). London, England: Sage. doi:10.1111/0022-4537.00239 doi:10.4135/9781446200919.n15 Rudman, L. A., Greenwald, A. G., & McGhee, D. E. (2001). Implicit Newcomb, T. M. (1981). Heiderian balance as a group phenomenon. self-concept and evaluative implicit gender stereotypes: Self and ingroup Journal of Personality and Social Psychology, 40, 862–867. doi: share desirable traits. Personality and Social Psychology Bulletin, 27, 10.1037/0022-3514.40.5.862 1164–1178. doi:10.1177/0146167201279009 Penton-Voak, I. S., Cahill, S., Pound, N., Kempe, V., Schaeffler, S., & Scheib, J. E., Gangestad, S. W., & Thronhill, R. (1999). Facial attractive- Schaeffler, F. (2007). Male facial attractiveness, perceived personality, ness, symmetry and cues of good genes. Proceedings of the Royal and child-directed speech. Evolution and Human Behavior, 28, 253–259. Society B, Biological Sciences, 266, 1913–1917. doi:10.1098/rspb.1999 doi:10.1016/j.evolhumbehav.2007.03.002 .0866 Perrett, D. I., Lee, K. J., Penton-Voak, I., Rowland, D., Yoshikawa, S., Schiappa, E., Gregg, P. B., & Hewes, D. E. (2005). The parasocial contact Burt, D. M., Henzi, S. P., Castles, D. L., & Akamatsu, S. (1998). Effects hypothesis. Communication Monographs, 72, 92–115. doi:10.1080/ of sexual dimorphism on facial attractiveness. Nature, 394, 884–887. 0363775052000342544 doi:10.1038/29772 Schilt, K., & Westbrook, L. (2009). Doing gender, doing heteronormativ- Perrett, D. I., May, K. A., & Yoshikawa, S. (1994). Facial shape and ity: “Gender normals,” transgender people, and the social maintenance judgments of female attractiveness. Nature, 368, 239–242. doi:10.1038/ of heterosexuality. Gender & Society, 23, 440–464. doi:10.1177/ 368239a0 0891243209340034 Perry, E. (1994). A prototype matching approach to understanding the role Sczesny, S., Spreemann, S., & Stahlberg, D. (2006). Masculine ϭ com- of applicant gender and age in the evaluation of job applicants. Journal petent? Physical appearance and sex as sources of gender-stereotypic of Applied Social Psychology, 24, 1433–1473. doi:10.1111/j.1559-1816 attributions. Swiss Journal of Psychology, 65, 15–23. doi:10.1024/1421- .1994.tb01558.x 0185.65.1.15 Pettigrew, T. F. (1998). Intergroup contact theory. Annual Review of Snyder, M., & Stukas, A. A., Jr. (1999). Interpersonal processes: The Psychology, 49, 65–85. doi:10.1146/annurev.psych.49.1.65 interplay of cognitive, motivational, and behavioral activities in social Pettigrew, T. F., & Tropp, L. R. (2006). A meta-analytic test of intergroup interaction. Annual Review of Psychology, 50, 273–303. doi:10.1146/ contact theory. Journal of Personality and Social Psychology, 90, 751– annurev.psych.50.1.273 783. doi:10.1037/0022-3514.90.5.751 Spence, J. T., Helmreich, R., & Stapp, J. (1975). Ratings of self and peers Posner, M. I., & Keele, S. W. (1968). On the genesis of abstract ideas. on sex-role attributes and their relation to self-esteem and conceptions of Journal of Experimental Psychology, 77, 353–363. doi:10.1037/ masculinity and femininity. Journal of Personality and Social Psychol- h0025953 ogy, 32, 29–39. doi:10.1037/h0076857 Prentice, D. A., & Carranza, E. (2002). What women and men should be, Swain, J. (2006). Reflections on patterns of masculinity in school settings. shouldn’t be, are allowed to be, and don’t have to be: The contents of Men and Masculinities, 8, 331–349. doi:10.1177/1097184X05282203 prescriptive gender stereotypes. Psychology of Women Quarterly, 26, Symons, D. (1995). Beauty is in the adaptations of the beholder: The 269–281. doi:10.1111/1471-6402.t01-1-00066 evolutionary psychology of human female sexual attractiveness. In P. R. Principe, C. P., & Langlois, J. H. (2012). Shifting the prototype: Experi- ence with faces influences affective and attractiveness preferences. So- Abramson, & S. D. Pinkerton (Eds.), Sexual nature/sexual culture (pp. cial Cognition, 30, 109–120. doi:10.1521/soco.2012.30.1.109 80–118). Chicago, IL: University of Chicago Press. Purkinje, J. (1825). Beobachtungen und Versuche zur Physiologie der Theodoridou, A., Rowe, A. C., Rogers, P. J., & Penton-Voak, I. S. (2011). Sinne. Neue Beitrage zur Kenntniss des Sehens in subjektiver Hinsicht Oxytocin administration leads to a preference for masculinized male [Observations and experiments on the physiology of the senses: New faces. Psychoneuroendocrinology, 36, 1257–1260. doi:10.1016/j contributions to the knowledge of seeing a subjective point]. Berlin, .psyneuen.2011.02.004 Germany: Reimer. Thornhill, R., & Gangestad, S. W. (1996). The evolution of human sexu- Rhodes, G. (2006). The evolutionary psychology of facial beauty. Annual ality. Trends in Ecology & Evolution, 11, 98–102. Review of Psychology, 57, 199–226. doi:10.1146/annurev.psych.57 Valenzano, D. R., Mennucci, A., Tartarelli, G., & Cellerino, A. (2006). .102904.190208 Shape analysis of female facial attractiveness. Vision Research, 46, This document is copyrighted by the American Psychological Association or one of its allied publishers. Rhodes, G., Halberstadt, J., Jeffery, L., & Palermo, R. (2005). The attrac- 1282–1291. doi:10.1016/j.visres.2005.10.024 This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. tiveness of average faces is not a generalized mere exposure effect. Webster, M. A., Kaping, D., Mizokami, Y., & Duhamel, P. (2004). Social Cognition, 23, 205–217. doi:10.1521/soco.2005.23.3.205 Adaptation to natural face categories. Nature, 428, 557–561. doi: Rhodes, G., Hickford, C., & Jeffery, L. (2000). Sex-typicality and attrac- 10.1038/nature02420 tiveness: Are supermale and superfemale faces super-attractive? British Webster, M. A., Werner, J. S., & Field, D. J. (2005). Adaptation and the Journal of Psychology, 91, 125–140. doi:10.1348/000712600161718 phenomenology of perception. In C. W. G. Clifford & G. Rhodes (Eds.), Rhodes, G., Jeffery, L., Clifford, C. W. G., & Leopold, D. A. (2007). The Fitting the mind to the world: Adaptation and after-effects in high-level timecourse of higher-level face aftereffects. Vision Research, 47, 2291– vision (pp. 241–277). Oxford, England: Oxford University Press. doi: 2296. doi:10.1016/j.visres.2007.05.012 10.1093/acprof:oso/9780198529699.003.0010 Rhodes, G., Jeffery, L., Watson, T. L., Clifford, C. W. G., & Nakayama, K. Williams, G. (2000). Don’t try to adjust your television – I’m Black: (2003). Fitting the mind to the world: Face adaptation and attractiveness Ruminations on the recurrent controversy over Whiteness on TV. Jour- aftereffects. Psychological Science, 14, 558–566. doi:10.1046/j.0956- nal of Gender, Race, and Justice, 4, 99–136. 7976.2003.psci_1465.x Willis, J., & Todorov, A. (2006). First impressions: Making up your mind Rhodes, G., Robbins, R., Jacquet, E., McKone, E., Jeffery, L., & Clifford, after a 100-ms exposure to a face. Psychological Science, 17, 592–598. C. W. G. (2005). Adaptation and face perception: How aftereffects doi:10.1111/j.1467-9280.2006.01750.x 1276 LICK AND JOHNSON

Wright, S. C., Aron, A., McLaughlin-Volpe, T., & Ropp, S. A. (1997). The Zajonc, R. B. (2001). Mere exposure: A gateway to the subliminal. Current extended contact effect: Knowledge of cross-group friendships and Directions in Psychological Science, 10, 224–228. doi:10.1111/1467- prejudice. Journal of Personality and Social Psychology, 73, 73–90. 8721.00154 doi:10.1037/0022-3514.73.1.73 Yamashita, J. A., Hardy, J. L., De Valois, K. K., & Webster, M. A. (2005). Stimulus selectivity of figural aftereffects for faces. Journal of Experi- mental Psychology: Human Perception and Performance, 31, 420–437. doi:10.1037/0096-1523.31.3.420 Received December 2, 2012 Zajonc, R. B. (1968). Attitudinal effects of mere exposure. Journal of Revision received August 19, 2013 Personality and Social Psychology, 9, 1–27. doi:10.1037/h0025848 Accepted August 26, 2013 Ⅲ

Members of Underrepresented Groups: Reviewers for Journal Manuscripts Wanted

If you are interested in reviewing manuscripts for APA journals, the APA Publications and Communications Board would like to invite your participation. Manuscript reviewers are vital to the publications process. As a reviewer, you would gain valuable experience in publishing. The P&C Board is particularly interested in encouraging members of underrepresented groups to participate more in this process.

If you are interested in reviewing manuscripts, please write APA Journals at [email protected]. Please note the following important points:

• To be selected as a reviewer, you must have published articles in peer-reviewed journals. The experience of publishing provides a reviewer with the basis for preparing a thorough, objective review.

• To be selected, it is critical to be a regular reader of the five to six empirical journals that are most central to the area or journal for which you would like to review. Current knowledge of recently published research provides a reviewer with the knowledge base to evaluate a new submission within the context of existing research.

• To select the appropriate reviewers for each manuscript, the editor needs detailed information. Please include with your letter your vita. In the letter, please identify which APA journal(s) you are interested in, and describe your area of expertise. Be as specific as possible. For example, “social psychology” is not sufficient—you would need to specify “social cognition” or “attitude change” as well. This document is copyrighted by the American Psychological Association or one of its allied publishers.

This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. • Reviewing a manuscript takes time (1–4 hours per manuscript reviewed). If you are selected to review a manuscript, be prepared to invest the necessary time to evaluate the manuscript thoroughly.